Tt

MC1

R

sgmats(ftc(cotwtcflb

toptsitfibci

lt5

Biochemical and Biophysical Research Communications 262, 638–642 (1999)

Article ID bbrc.1999.1251, available online at http://www.idealibrary.com on

0CA

argeting of Green Fluorescent Protein Expressiono the Cell Surface

aria Simonova, Ralph Weissleder, Nikolai Sergeyev, Natalia Vilissova, and Alexei Bogdanov, Jr.1

enter for Molecular Imaging Research, Massachusetts General Hospital, Building 149,3th Street, Charlestown, Massachusetts 02129

eceived July 2, 1999

tGv(dititp(tMGl

tgptiNpbadptae

btmadtcc(C

We have previously reported on GPI-anchored fu-ion proteins that bind radioactive isotopes. We tar-eted their expression to the cell surface to obtain aarker protein detectable by nuclear and optical im-

ging (1, 2). Here we suggest a novel approach forargeting a model protein (GFP) to the exoplasmicurface of the plasma membrane. An expression vectorpcPEP-GFP) was constructed containing GFP cDNAused with the fragment encoding the N-terminal cy-oplasmic domain and signal peptide/membrane an-horing domain of the rabbit neutral endopeptidasePEP-GFP). Flow cytometry showed green fluores-ence in 45% of cells transfected with GFP and in 34%f cells transfected with PEP-GFP (24 h after transfec-ion). Fluorescence microscopy of fixed cells stainedith rhodaminated anti-GFP antibodies showed posi-

ive reaction only in the case of PEP-GFP-transfectedells indicating cell-surface expression. The PEP-GFPusion protein was identified as a component of theight microsomal and Golgi fractions by immuno-lotting. © 1999 Academic Press

Green fluorescent protein (GFP) from Aequeorea vic-oria (3, 4) can be expressed in a variety of livingrganisms as fusion proteins for tracking individualroteins and to study cellular dynamics. Mutationshat were introduced in the chromophore-generatingequence of wild-type GFP (e.g., Ser65 to Thr) havencreased the quantum yield of GFP fluorescence andhe fluorescence excitation maximum has been shiftedrom 395 nm to 490 nm (5, 6). The utility of GFP forntravital microscopy has furthermore been broadenedy additional mutations introduced to optimize theodon usage of GFP in mammalian cells (7, 8), and toncrease GFP solubility by altering its folding charac-

1 To whom correspondence should be addressed at Center for Mo-ecular Imaging Research, Rm. 5420, Massachusetts General Hospi-al, Bldg. 149, 13th Street, Charlestown, MA 02129. Fax: (617) 726-708. E-mail: abogdanov@helix.mgh.harvard.edu.

638006-291X/99 $30.00opyright © 1999 by Academic Pressll rights of reproduction in any form reserved.

eristics (9, 10). Expression of membrane-associatedFP-tagged proteins has been utilized for selectivelyisualizing proteins involved in intercellular adhesion11) and in research of peripheral membrane proteinynamics (12). GFP has also been used to visualizentracellular vesicle trafficking and transport events inhe endoplasmic reticulum and Golgi (13, 14), traffick-ng to the plasma membrane (15) and in investigatinghe diffusional mobility of TGN38 proteins (16). Ex-ression of GFP fusions with b-galactosyltransferase17) or spectrin (18) has facilitated cell cycle analysis ofransiently transfected cells using flow cytometry (18).oreover, double labeling with spectral mutants ofFP has allowed the simultaneous study of two popu-

ations of cell proteins (17, 19, 20).The expression of GFP on the exoplasmic leaflet of

he plasma membrane has been attempted before. Alycosylphosphatidylinositol (GPI)-anchored fusionrotein has been suggested by De Angelis et al. as aool in investigating plasma membrane protein dynam-cs (21). This strategy requires an additional-terminal leader peptide for directing the nascentolypeptide through endoplasmic reticulum mem-rane. In addition, the cell-surface expression of GPI-nchored membrane protein may substantially vary inifferent cell types since it requires several steps ofost-translational modification including cleavage ofhe C-terminal sequence and selective trans-midination with GPI precursor which is assembled inndoplasmic reticulum and Golgi (22).We have previously developed GFP fusion proteins

inding radioactive isotopes and have attempted toarget their expression to the cell surface to obtain aarker protein detectable by nuclear and optical im-

ging (1, 2). The goal of the current study was toevelop and characterize a novel plasma membrane-argeted expression system utilizing the N-terminalytoplasmic domain and signal peptide/membrane an-horing domain of rabbit neutral endopeptidase 24.11enkephalinase). Enkephalinase has a large catalytic-terminal domain exposed on the exoplasmic side of

tcftg

M

ie(eap

gr(TaCpBCSfTpMmnATKlKceu

CMs7efM

cfWp1itl1sA(a

fwo

between incubations were accomplished by centrifuging cellstFCta

scssthAuawFtaiPf

mitPbspas

R

vTHTidmpspsvlr

FsPGG1Fna4

Vol. 262, No. 3, 1999 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

he plasma membrane as a result of N-terminal an-horing to the membrane (23). We hypothesized thatusing N-terminal fragment of type II membrane pro-ein with the protein of interest may provide a simpleeneric strategy for cell surface targeting.

ATERIALS AND METHODS

Expression vectors. The pcPEP-GFP vector was constructed us-ng a synthetic S65T mutant gfp sequence with optimized codons fornhanced expression in eukaryotes (7). The original plasmidphGFPS65T, kindly provided by Dr. B. Seed, Massachusetts Gen-ral Hospital) was digested with HindIII-NotI restriction endonucle-ses and 0.7 kb gfp fragment was cloned into HindIII-NotI digestedcDNA3 vector (Invitrogen, Carlsbad, CA).To construct pcPEP-GFP vector, gfp sequence was amplified to-

ether with a 36-bp HindIII-BamHI adapter using polymerase chaineaction (PCR). PCR was performed using Pwo DNA polymeraseDNA amplification kit; Boehringer-Mannheim, Indianapolis, IN).he sense primer was: 59-GGAAGCTTGAATTCGTGAGCAAGGG-39nd antisense primer: 59-CAGGATCCCTCCTCCACATGGACAT-CTCCTTCAAGCTTGTACAGCTCGTCCATGCCG-39). The PCRroduct was digested with EcoRI-BamHI and subcloned into EcoRI-amHI sites of pBlueScriptSKII (1) vector (Stratagene, La Jolla,A) to generate aTAG stop signal in resultant pBSKS-GFP plasmid.elected clones were first digested with EcoRI. Using the Klenow

ragment the EcoRI restriction site was filled to obtain a blunt end.he plasmid was then digested with KpnI for further cloning. TheSVENK19 vector (a generous gift of Dr. Guy Boileau, University ofontreal) was used for PCR to amplify the cytoplasmic domain andembrane anchoring domains, together with 40-bp adapter of rabbiteutral endopeptidase 24.11 (sense: 59-GCAAGATTTGGTACC-TGGGAAGATCAG-39 and antisense: 59-GTCGAGCAGCTGATTT-ATGCAGTCT-39-primers. The restriction sites are underlined andozak sequence surrounding ATG start codon is marked with bold

etters). The fragment was digested with KpnI-PvuII and cloned intopnI-EcoRI (blunt ended)-digested pBSKS-GFP plasmid. Selected

lones were digested with the KpnI-XbaI and cloned into pcDNA3xpression vector. Clones with expected 0.9 kb insert were selectedsing KpnI-XbaI restriction analysis and analyzed by sequencing.

Cell transfection. COS-1 cells were obtained from ATCC (1650-RL, Rockville, MD) and maintained in DMEM medium (Cellgro,ediatech, Washington, DC) supplemented with 10% fetal bovine

erum (FBS, HyClone Laboratories, Logan, UT). Cells were plated at0% confluency in 12 well plates or on glass coverslips (Fisher Sci-ntific, Pittsburgh, PA) 24 h before transfection. Cells were trans-ected for 4 h using TransIT-100 reagent (PanVera Corporation,

adison, WI), as described by the manufacturer.

Immunofluorescence analysis of transiently transfected COS-1ells. At 24 and 48 h after transfection, cells were fixed with 1%ormaldehyde in HBSS, rinsed with 10% horse serum (Mediatech,

ashington, DC) in PBS (pH 7.4) and incubated for 1h with rabbitolyclonal anti-GFP antibody (Clontech, Palo Alto, CA) diluted at:1000 in the incubation buffer (10% horse serum in PBS). Followingncubation, cells were rinsed with wash buffer and incubated sequen-ially with biotinylated goat anti-rabbit antibody and rhodamine-abeled streptavidin (Pierce, Rockford, IL) diluted at 1:4,000 and:500, respectively. Control experiments were performed in the ab-ence of the primary antibody. Cells were examined using a Zeissxiovert 100TV (Wetzlar, Germany) equipped with a CCD camera

Photometrics, Tuscon, AZ). Two channel detection using rhodaminend fluorescein filters was performed.

Flow cytometry. Transfected and control cells were dislodgedrom the substrate using 1 mM EDTA in ice-cold PBS and washedith GM-saline. Cells were fixed, incubated with primary and sec-ndary antibodies and streptavidin as described above. Washings

639

hrough a step of 40% Hystopaque-1077 in HBSS (twice, 4°C, 800 g).low cytometry analysis was performed using FACSCalibur andellQuest Software (Beckton Dickinson Immunocytochemistry Sys-

ems, San Jose, CA). The FL1 (530/30) bandpass for GFP detectionnd FL2 (572/26) bandpass for rhodamine detection were used.

Subcellular fractionation. Transfected COS-1 were grown as de-cribed above and harvested at near confluency by scraping in ice-old PBS, 0.5 mM EDTA, pH 7.4. Four to 8 mln cells were washed byedimentation and resuspended in 2 mL 10 mM Hepes, 0.25 Mucrose, 1 mM EDTA pH 7.4 (HSE). Cells were homogenized using aight-fitting Dounce homogenizer. Post-nuclear supernatant of cellomogenate was adjusted to 35% of Iodixanol (OptiPrep, Nycomed).discontinuous 10 mL gradient of Iodixanol (10–30% in HSE) was

nderlayered using 1.5 mL of cell homogenate and centrifuged usingSW41Ti rotor at 28,500 rpm for 1.5 h at 4°C. The gradient fractionsere analyzed for the presence of specific enzymatic markers (24).luorescence (lem490/lex508 nm) was measured in individual frac-ions, fractions containing microsomes, Golgi and cytosol were poolednd protein concentration measured. Proteins were analyzed bymmunoblotting using anti-GFP polyclonal antibody (Clontech,alo Alto, CA) and anti-rabbit-peroxidase conjugate (Pierce, Rock-

ord, IL).

Immunoblotting. Transfected and control cells were lysed in 200l of 1% Igepal, 0.2% SDS in 50 mM Tris, pH 7 and Complete

nhibitors (Boehringer-Mannheim, Ridgefield, CT). Lysate was cen-rifuged at 12,000 g for 5 min. Proteins were resolved using 10%AGE, transferred to PVDF membranes and analyzed by immuno-lotting using anti-GFP antibodies (Clontech, dilution 1:3,000) andecondary biotinylated antibodies (dilution 1:2,000) and avidin-eroxidase (Bio-Rad, dilution 1:2,000). Peroxidase activity was visu-lized using DAB/a-chloronaphthol mixture and digitized using acanner.

ESULTS

Expression vector. To construct the pcPEP-GFPector, gfp cDNA was amplified using PCR method.he reverse primer was designed to generate aindIII-BamHI 36-bp adapter at 39-terminus of gfp.he PCR fragment encoding for the 27-amino acid res-

dues of the positively charged N-terminal cytoplasmicomain and the 23-residue hydrophobic signal peptide/embrane anchoring domain of rabbit neutral endo-

eptidase 24.11 (23) was inserted upstream to the gfpequence into a pBlueScriptSKII(1) vector. The com-lete fragment was then subcloned into the KpnI-XbaIites of pcDNA3 (Fig. 1). The resultant pcPEP-GFPector thus encoded for a gfp as an integral extracel-ular fusion type II protein, due to the N-terminalegion of rabbit neutral endopeptidase 24.11.

Extracellular and intracellular expression of GFP.luorescence microscopy and FACS analysis of tran-iently transfected cells confirmed the expression ofEP-GFP fusion protein and also of control cytosolicFP. The number of GFP positive cells and levels ofFP fluorescence after transfection are shown in Tablefor both PEP-GFP and control GFP-transfected cells.low cytometry of fixed cells showed that at 48 h theumber of GFP-positive cells was 1.5-fold higher thant 24 h after transfection (in the case of PEP-GFP). At8 h after transfection the mean fluorescence of GFP-

tthtsmfiIrbicct(i2cGpdl

tttSmtpwltp

D

vhnipmdescoPtflckGattcgaptc

LE

CGP

Vol. 262, No. 3, 1999 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

ransfected cells was 3-4 fold higher than PEP-GFPransfected cells. Approximately 50% of cells exhibitedigh level of fluorescence (20 fold higher than non-ransfected cells; Table 1). To differentiate between cellurface and cytoplasmic fluorescence, cells were imper-eabilized for immunoglobulins by formaledhydexation and probed with a polyclonal rabbit anti-GFPgG followed by anti-rabbit biotinylated IgG andhodamine-streptavidin. The experiment revealedright punctate staining of the plasma membrane vis-ble in the red (rhodamine) channel (Figs. 2A, 2B). Noell surface fluorescence could be detected in controlells transfected with the pcGFP vector (Fig. 2D)hough intracellular green fluorescence was presentFig. 2F). At 24 h, fusion protein fluorescence was seenn the perinuclear space of many transfected cells (Fig.D). The typical asymmetric distribution of fluores-ence surrounding the nucleus indicates that PEP-FP was concentrating in the Golgi compartment. Thisarticular pattern of protein localization could not beetected at 48 h after transfection because of highevels of cytoplasmic fluorescence. Immunoblotting of

FIG. 1. The scheme of the pcPEP-GFP expression vector.

TAB

Transfection Efficiency and Levels of GFP Ex

Transfectedcell type

T

24

Mean FL1intensity

% FL1positive* p

ontrol (mock) 7.8 1.0 0FP 861 45.6 1EP-GFP 169 34.3 2

* % GFP-positive cells, i.e., transfection efficiency.# % FL1-positive cells with surface-bound anti-GFP antibody, mea

640

otal cell lysates of PEP-GFP expressing cells showedhat approximately 30% of total anti-GFP positive pro-ein consisted of a 38 kD PEP-GFP fusion (Fig. 3).ubcellular fractionation of cells showed that microso-al and Golgi compartments contained the major frac-

ion of expressed PEP-GFP. However, the latter com-artment contained two protein expression productsith molecular weights of 38 and 32 kD. The full-

ength PEP-GFP was essentially absent from the cy-osol that contained only the 32 kD anti-GFP positiveroduct.

ISCUSSION

We designed and tested a pcPEP-GFP expressionector encoding a small N-terminal cytoplasmic andydrophobic membrane-spanning region of the rabbiteutral endopeptidase 24.11 (enkephalinase, type II

ntegral membrane protein (23) for targeting of a modelrotein, GFP, to the exoplasmic surface of plasmaembrane. This new approach of protein expression

oes not require post-translational modification. Alllements responsible for polypeptide targeting to theecretory pathway and membrane anchoring are en-oded within the contiguous N-terminal sequence. Webserved a relatively high level of cell transfection withEP-GFP fusion (34% cells were fluorescent 24 h afterransfection and 50% at 48 h). The levels of PEP-GFPuorescence were somewhat lower than GFP fluores-ence that may reflect the difference in protein foldinginetics. As shown by fluorescence microscopy, PEP-FP fusion was expressed at high levels (Figs. 2D, 2E)nd transiently accumulated in Golgi at 24 h afterransfection (Fig. 2D). However, anti-GFP reactive pro-ein was simultaneously identified on the surface ofells transfected with PEP-GFP (Fig. 2A). The most ofreen fluorescent cells transfected with PEP-GFP re-cted with the polyclonal anti-GFP antibody as op-osed to control transfectants (Figs. 2C, 2F). Distribu-ion of PEP-GFP on the cell surface appears to belustered (Figs. 2A, 2B). The size of the fluorescent

1

ssion on the Cell Surface by Flow Cytometry

after transfection, hours

48

L2tive#

Mean FL1intensity

% FL1positive*

% FL2positive#

0.0 9.0 0.8 0.2 6 0.10.0 1483 78.6 1.1 6 0.00.2 519.5 49.6 5.9 6 0.4

SD.

pre

ime

% Fosi

.1 6

.2 6

.1 6

n 6

irpptitnmfsflct(ieb(afs

Npem

uTcm

A

a

R

1

1

1

1

111

t

lfs

Vol. 262, No. 3, 1999 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

mmunocomplexes on the surface of the cells does noteflect the true size of the cell-surface expressed fusionrotein. This is due to the ability of expressed fusionrotein to bind several secondary antibodies that, inurn, bind labeled streptavidin molecules. However, its also possible that punctuate distribution may reflecthe tendency of GFP to polymerize and to formanoclusters. High levels of transiently expressedembrane-associated GFP may potentially provide a

avorable environment for multimerisation of the fu-ion protein (6, 21). In 6% of cells exhibiting GFPuorescence the number of PEP-GFP molecules on theell surface was high enough for these cells to be de-ected by flow cytometry using anti-GFP antibodiesTable 1). Analysis of PEP-GFP fusion protein variantsn different cellular compartments showed the pres-nce of the full-length 38 kD protein and additionaland corresponding to the cytosolic 32 kD fragmentFig. 3). The mass of this fragment is higher than GFPlone (29 kD) which suggests a site susceptibleor proteolysis located within cytoplasmic/membranepanning domain of endopeptidase.In conclusion, we demonstrated that by using-terminal fragment of endopeptidase for anchoring tolasma membrane, corresponding fusion proteins can bexpressed at the cell surface of eukaryotic cells. Further-ore, we characterized expression of GFP-fusion prod-

FIG. 2. Immunofluorescence images of COS-1 cells transiently transfection) and GFP control (C, F). A, B, C: rhodamine (red) chan

FIG. 3. Immunoblotting of the complete lysate (T) and subcellu-ar fractions (M, microsomal fraction; G, Golgi fraction; C, cytosolicraction) obtained from pcPEP-GFP-transfected cells. GFP control ishown at the right.

641

cts by flow cytometry and subcellular fractionation.hese findings are relevant for understanding the pro-ess of type II membrane protein transport to plasmaembrane as well as for receptor engineering.

CKNOWLEDGMENTS

This study was supported in part by NIH 5RO1 NS35258-03 (R.W.)nd 5RO1 CA74424-01 (A.B.).

EFERENCES

1. Bogdanov, A. A., Jr., Simonova, M., and Weissleder, R. (1997)Nature Biotechnol. Short Rep. 8, 22.

2. Bogdanov, A. A., Jr., and Weissleder, R. (1998) Trends Biotech.16, 5–10.

3. Prasher, D. C., Eckenrode, V. K., Ward, W. W., Prendergast,F. G., and Cormier, M. J. (1992) Gene 111, 229–233.

4. Chalfie, M., Tu, Y., Euskirchen, G., Ward, W. W., and Prasher,D.C. (1994) Science 263, 802–805.

5. Heim, R., Cubitt, A. B., and Tsien, R. Y. (1995) Nature 373,663–664.

6. Cubitt, A. B., Heim, R., Adams, S. R., Boyd, A. E., Gross, L. A.,and Tsien, R. Y. (1995) Trends Biochem. Sci. 20, 448–455.

7. Haas, J., Park, E. C., and Seed, B. (1996) Curr. Biol. 6, 315–24.8. Zolotukhin, S., Potter, M., Hauswirth, W. W., Guy, J., and Muzy-

czka, N. (1996) J. Virol. 70, 4646–4654.9. Cormack, B. P., Valdivia, R. H., and Falkow, S. (1996) Gene 173,

33–38.0. Crameri, A., Whitehorn, E. A., Tate, E., and Stemmer, W. P. C.

(1996) Nature Biotechnol. 14, 315–319.1. Adams, C. L., Chen, Y. T., Smith, S. J., and Nelson, W. J. (1998)

J. Cell Biol. 142, 1105–1119.2. Yokoe, H. and Meyer, T. (1996) Nature Biotechnol. 14, 1252–

1256.3. Presley, J., Cole, N., Schroer, K., Hirschberg, K., Zaal, K., and

Lippincott-Swartz, J. (1997) Nature 389.4. Girotti, M., and Banting, G. (1996) J. Cell Sci. 109, 2915–2926.5. Horn, M., and Banting, G. (1994) Biochem. J. 301, 69–73.6. Cole, N. B., Smith, C. L., Sciaky, N., Terasaki, M., Edidin, M.,

and Lippincott-Schwartz, J. (1996) Science 273, 797–801.

sfected with PEP-GFP fusion protein (A, D: 24 h; B, E: 48 h afterD, E, F: GFP (green) channel.

rannel;

17. Ellenberg, J., Lippincott-Schwartz, J., and Presley, J. F. (1998)

1

1

2

21. De Angelis, D. A., Miesenbock, G., Zemelman, B. V., and Roth-

22

2

Vol. 262, No. 3, 1999 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

Biotechniques 25, 838–846.8. Kalejta, R. F., Shenk, T., and Beavis, A. J. (1997) Cytometry 29,

286–91.9. Rizzuto, R., Brini, M., De, G., Rossi, R., Heim, R., Tsien, R., and

Pozzan, T. (1996) Curr. Biol. 6, 183–188.0. Zimmermann, T., and Siegert, F. (1998) Biotechniques 24, 458–

461.

642

man, J. E. (1998) Proc. Natl. Acad. Sci. USA 95, 12312–12316.2. Cross, G. A. (1990) Ann. Rev. Cell Biol. 6, 1–39.3. Roy, P., Chatellard, C., Lemay, G., Crine, P., and Boileau, G.

(1993) J. Biol. Chem. 268, 2699–704.4. Spector, D., Goldman, R., and Leinwand, L. (1998) Cells: Culture

and Biochemical Analysis of Cells, Vol. 1, Cold Spring HarborLaboratory Press, Cold Spring Harbor, NY.